Fracking With Numbers

A study published this week in the Proceedings of the National Academy of Sciences is looking to fact check the EPA’s methane emissions estimates for fracking in natural gas wells. The study looked at 190 natural gas sites to see how much methane actually managed to escape throughout the process. It surveyed 150 production sites, 27 well completion flowbacks, 9 well unloadings, and 4 workovers (although the workovers ended up contributing so little their data was not included in the published work). They found that the EPA’s estimates were fairly close to their own findings, with a few notable exceptions.

The EPA’s numbers are based on engineering estimates for methane loss during mining along with average emission factors developed back in the ‘90s. Given that extraction technology has progressed significantly since then, these emissions rates may no longer be accurate. To try and test them, this group of researchers took to the field, visiting the actual natural gas sites and taking direct emissions measurements as the mines operated.

A diagram showing how fracking works. It’s a bit complicated and involved, but it’s very good, and fairly objective in its presentation (Source:

First off, here are some quick fracking basics with some helpful terminology. Fracking works by digging a hole towards some underground source of natural gas that is trapped in rock, typically shale. Once the hole is dug, they blast that source with a mixture of water and sand along with other additives to fracture it and thus release the gas. Before they can extract it though, they have to pull all that water and sand and whatnot back out. That’s what a well completion flowback is, when that fracturing mixture literally flows back out. While there are mechanisms in place to capture whatever gases may come with it, often gases manage to escape and so flowbacks are still a major source of methane emissions. Now during the actual extraction process, sometimes the pipes get clogged with water or other liquid hydrocarbons that the well has produced, which will obviously impede the flow of your gas. As such, the pipes need to be occasionally cleared out. That’s what well unloading is. Depending on the rate of flow for natural gas here, these can be serious problems for methane escaping. Workovers are any invasive intervention in the well process, but since they aren’t really important here, I’m not going to go into detail.

An example of a manifold used in flowback operations (Source:

Now for the group’s results. With regards to flowback, they found a significant overestimation on the part of the EPA’s inventory. For the 27 flowback events they monitored at sites on the Gulf Coast, in the Appalachians, in the Rockies, and across the middle of the US, they found an average flowback emission of 1.7 megagrams of methane per event. The EPA’s estimate was 81 megagrams per event. Yes that difference is really as ridiculously large as it seems. The reason is pretty simple too. Each well has a potential flowback emission, which is just all of the gas that could be released in the flowback process. The averages for potential flowback for both this study and the EPA’s were close, 124 Mg for the study and 151 Mg for the EPA. The difference is that for their estimates, the EPA assumes that about half of that potential will be captured, while the other half escapes (that 81 Mg from before). In contrast, these measurements suggest that about 98% of the potential emissions are stopped.

Unloading processes had the same problem with overestimation. For the nine events surveyed, the EPA’s methods for estimation predicted 5.2 Mg of methane emissions per event. The actual measurements only showed 1.1 Mg per unloading, a fivefold decrease from the EPA estimate. Here the problem is a bit more difficult to address. A survey from the American Petroleum Institute and America’s Natural Gas Alliance was used to design the method for the EPA’s estimates. According to this survey, of 2,901 unloadings, 100 account for 50% of total emissions. That means only 3% of the surveyed sites contributed half the total emissions. If this group’s survey didn’t cover that 3%, then there’s no way their estimates could line up with the EPA’s. That being said, what the hell is with those 100 sites? Somebody should really look into that. That’s ridiculous.

Just your average, Midwestern, picturesque fracking site (Source:

For the production site survey, the group focused on differences in equipment leaks (from use of separators, pneumatic controllers, and other specialized means of gas control). It is difficult to compare their results to the EPA’s estimates, because the EPA does not assign specific emission reduction values to any of these individual pieces of equipment. Still, there were some striking differences. For example, when looking at low bleed and intermittent bleed pneumatic controllers, they found that their emissions were 29% and 270% higher than the EPA’s predicted factors, respectively. So that’s pretty impressive.

Now overall, how do these differences affect the EPA’s grand estimate? The truth is, not a whole lot.  The study’s total estimate for national emissions came out at about 2,300 gigagrams of methane, while the EPA sat around 2,545 gigagrams. The only problem with that result is that it is still heavily reliant on EPA estimates. The group took its results and estimated the national levels of emissions from flowbacks, unloadings, and equipment leaks to be around 957 Gg (about 300 Gg less than the EPA’s own estimate). They then simply added this to the EPA’s estimates for emissions from other sources. Personally, given they had just shown the EPA’s estimates are a bit flawed, I think that seems a bit silly. What this study really demonstrates is that we need much more detailed and thorough information about fracking and its emissions (this is of course also ignoring all the emissions that mange to get into the local water table, check out literally any news story about fracking ever, or the Gasland documentaries). Estimates from engineering and abstraction are functional, but not nearly as trustworthy as estimates from direct readings. So get to work science.

Original article:


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s